targets

Output features to be optimized

Functions

resolve_target_names(targets, target_names, *)	Resolve a name filter against a list of ``Target``s.
to_tensor(f[, dtype, device])	Validate and convert f to a torch.Tensor.

Classes

Target(name[, f_transform, aim, ...])

Base class for optimization response targets.

class obsidian.parameters.targets.Target(name: str, f_transform: str | None = 'Standard', aim: str = 'max', tracking_only: bool = False, threshold: float | None = None)

Bases: object

Base class for optimization response targets.

Parameters:

• name – Name of the target/response variable

• f_transform – Transform function to apply (default: “Standard”)

• aim – Optimization direction - “max” or “min” (default: “max”)

• tracking_only – If True, target is tracked but not optimized (default: False)

• threshold – Optional threshold value for characterization tasks. - If aim=”max”: characterize regions where response >= threshold - If aim=”min”: characterize regions where response <= threshold

get_threshold(transformed: bool = True) → float | None

Get the threshold value, optionally transformed.

Parameters:

transformed – If True, apply the target’s transform to the threshold (default: True)

Returns:

The threshold value (transformed or raw), or None if no threshold is set

Raises:

UnfitError – If transformed=True but the transform function hasn’t been fit yet

classmethod load_state(obj_dict: dict)

Loads the state of the target object from a dictionary.

Parameters:

• cls (class) – The class of the target object.

• obj_dict (dict) – A dictionary containing the state of the target object.

Returns:

The loaded target object.

save_state() → dict

Saves the state of the object as a dictionary.

Returns:

A dictionary containing the state of the object.

Return type:

dict

transform_f(f: float | int | _SupportsArray[dtype[Any]] | _NestedSequence[_SupportsArray[dtype[Any]]] | bool | complex | str | bytes | _NestedSequence[bool | int | float | complex | str | bytes], inverse=False, fit=False)

Converts a raw response to an objective function value (“score”). Cost-penalization and response transformation should be handled here.

Parameters:

• f (array-like) – The column(s) containing the response values (y)

• inverse (bool, optional) – An indicator to perform the inverse transform. Defaults to False.

• fit (bool, optional) – An indicator to fit the properties of the transform function. Defaults to False.

Returns:

An array of transformed f values matching the responses in Z

Return type:

pd.Series

Raises:

• TypeError – If f is not numeric or array-like

• UnfitError – If the transform function is called without being fit first

obsidian.parameters.targets.resolve_target_names(targets: list, target_names: list[str] | None, *, require_thresholds: bool = False, drop_tracking_only: bool = True) → list

Resolve a name filter against a list of ``Target``s.

Centralizes the existence / tracking-only / threshold checks shared by plotting and characterization. None means “all targets that pass the filters”; an explicit list raises on unknown names so typos surface.

Parameters:

• targets – All targets defined on the campaign.

• target_names – Optional subset to restrict to.

• require_thresholds – If True, every selected target must have a threshold set (used by characterization / passfail / confidence).

• drop_tracking_only – If True, tracking_only targets are dropped when target_names is None and rejected when listed explicitly.

Returns:

Selected targets in the order they appear in targets.

obsidian.parameters.targets.to_tensor(f: Any, dtype: dtype = torch.float64, device: device | str | int | None = None) → Tensor

Validate and convert f to a torch.Tensor. Accepts: torch.Tensor, numpy.ndarray, pandas Series/DataFrame, Python scalar (int/float), or list/tuple of numerics.

Parameters:

• f – input to convert

• dtype – optional torch dtype for the resulting tensor

• device – optional torch device for the resulting tensor

Returns:

torch.Tensor